rowhanm / sampleCNN-pytorch

Pytorch implementation of "Sample-level Deep Convolutional Neural Networks for Music Auto-tagging Using Raw Waveforms"

Sample-level Deep CNN

Pytorch implementation of Sample-level Deep Convolutional Neural Networks for Music Auto-tagging Using Raw Waveforms

Data

MagnaTagATune Dataset

• Used tag annotations and audio data

Model

9 1D conv layers and input sample size of 59049 (~3 seconds)

Procedures

• Fix config.py file
• Data processing
  • run python audio_processor.py : audio (to read audio signal from mp3s and save as npy)
  • run python annot_processor.py : annotation (process redundant tags and select top N=50 tags)
    • this will create and save train/valid/test annotation files
• Training
  • You can set multigpu option by listing all the available devices
  • Ex. python main.py --gpus 0 1
  • Ex. python main.py will use 1 gpu if available as a default

Tag prediction

• run python eval_tags.py --gpus 0 1 --mp3_file "path/to/mp3file/to/predict.mp3"

References

• https://github.com/jongpillee/sampleCNN
• https://github.com/tae-jun/sample-cnn
• https://github.com/keunwoochoi/magnatagatune-list